Re: [PATCH] A better (?) _AC_EXEEXT

[Pavel Roskin]

Hello, Morten!

> * Akim
> | I agree there are *two* issues.  One is that there is a AC_REQUIRE
> | circular dependency because AC_EXEXT uses AC_LINK_IFELSE, the other
> | is that both AC_LINK_IFELSE and AC_TRY_EVAL(ac_link) are
> | inappropriate here.

This can be solved e.g. with _AC_LINK_IFELSE that is the same as
AC_LINK_IFELSE but accepting _possible_ suffix as argument.

> Aha. How about the attached patch, then?

> +  if (test conftest -ef conftest.exe) >/dev/null 2>&1; then

Why do you want to set ac_exeext to ".exe" is conftest is the same as
conftest.exe? If there is a problem with ac_exeext not being ".exe" try to
reproduce it here instead of relying on "-ef" which may (or may not) drive
old shells crazy.

There is no reason to prefer ".exe" over "" unless there are problems
with the later.

I.e. "cp conftest conftest.foo" may fail - try it here.

> +  # Couldn't use empty suffix, try with suffix commonly used
> +  # on MSWindows platforms.

It's an insult for OS/2 people. Use "DOS-based platforms" or "PC
platforms"

Also it appears (I haven't tested) that ac_exeext is not guaranteed to be
equal to ac_cv_exeext at the end of the macro.

Regards,
Pavel Roskin